Multiobjective Optimization of Silver-Nanowire Deposition for Flexible Transparent Conducting Electrodes

Area of Science:

• Materials Science and Engineering
• Nanotechnology
• Machine Learning Applications

Background:

• Transparent conducting electrodes (TCE) are crucial for electronic devices, requiring a balance between high optical transmittance and electrical conductance.
• Traditional optimization methods using a single figure of merit struggle to meet the distinct transmittance and conductance requirements of diverse applications.
• Spin coating of silver nanowires (AgNWs) is a promising technique for fabricating TCEs, but requires precise parameter control for optimal performance.

Purpose of the Study:

• To develop and apply machine learning (ML) models for optimizing the spin coating process of silver nanowires for TCE fabrication.
• To address the challenge of simultaneously achieving high transmittance and high conductance in TCEs, moving beyond single-metric optimization.
• To identify specific processing parameters that enable TCEs to meet demanding application requirements, such as transmittance ≥ 75% and sheet resistance ≤ 15 Ω/sq.

Main Methods:

• Utilized machine learning (ML) models to analyze the complex relationship between spin coating parameters and TCE performance.
• Performed Pareto front analysis on ML model predictions to identify optimal processing conditions that balance competing transmittance and conductance characteristics.
• Validated the ML-identified parameters by fabricating and characterizing TCEs under the predicted optimal conditions.

Main Results:

• Demonstrated that ML-guided optimization can effectively navigate the trade-off between transmittance and conductance in AgNW TCEs.
• Achieved target performance metrics of transmittance ≥ 75% and sheet resistance ≤ 15 Ω/sq, which were previously challenging to attain simultaneously.
• Identified specific spin coating parameters through ML analysis that lead to superior TCE performance.

Conclusions:

• Machine learning, particularly Pareto front analysis, provides a powerful tool for optimizing complex material fabrication processes like AgNW spin coating.
• The ML approach successfully overcomes the limitations of single-figure-of-merit optimization, enabling the tailored fabrication of TCEs for specific applications.
• This study validates the feasibility of achieving high-performance AgNW TCEs with desired optical and electrical properties using ML-driven process control.